All-or-none switchover to address split-brain problems in multi-chassis link aggregation groups

Abstract

Systems and methods utilize an all-or-none switchover to prevent split-brain problems in a Multi-Chassis Link Aggregation Group (MC-LAG) network. A standby node in the MC-LAG network can perform the steps of remaining in a standby state responsive to a loss of adjacency with an active node, wherein, in the standby state, all standby links between the standby node and a common endpoint are non-distributing; monitoring frames transmitted by the common endpoint to the standby node over the standby links; and determining based on the monitoring frames whether all active links between the active node and the common endpoint have failed and entering an active state with all the standby links distributing based thereon.

What is claimed is: 
     
       1. A method utilizing all-or-none switchover to prevent split-brain problems in a Multi-Chassis Link Aggregation Group (MC-LAG) network implemented by a standby node, the method comprising:
 remaining in a standby state responsive to a loss of adjacency with an active node, wherein, in the standby state, all standby links between the standby node and a common endpoint are non-distributing; 
 monitoring frames transmitted by the common endpoint to the standby node over the standby links; and 
 determining based on the monitoring frames whether all active links between the active node and the common endpoint have failed and entering an active state with all the standby links distributing based thereon. 
 
     
     
       2. The method of  claim 1 , further comprising:
 determining based on the monitoring frames whether less than all of the active links have failed and remaining in the standby state and continuing monitoring the frames transmitted by the common endpoint over the standby links based thereon. 
 
     
     
       3. The method of  claim 1 , wherein the monitoring checks for a presence of SYNC bits from the common endpoint with each SYNC bit set to TRUE indicative of a switch by the common endpoint of one of the active links to one of the standby links. 
     
     
       4. The method of  claim 1 , wherein the common endpoint is communicatively coupled to both the active node and the standby node in an active/standby triangle topology. 
     
     
       5. The method of  claim 1 , wherein the common endpoint is configured to operate Link Aggregation Control Protocol (LACP) and an N:N link-level redundancy between the active node and the standby node. 
     
     
       6. The method of  claim 1 , wherein the common endpoint is unaware the active node and the standby node are in separate network elements. 
     
     
       7. The method of  claim 1 , wherein the loss of adjacency with the active node is based on a failure or fault on a link between the active node and the standby node used for coordination of the active node and the standby node in a Redundant Group, while the active node and the standby node are both operational. 
     
     
       8. A standby node in a Multi-Chassis Link Aggregation Group (MC-LAG) network configured with all-or-none switchover to prevent split-brain problems, the standby node comprising:
 a plurality of ports in a logical Link Aggregation Group (LAG) with an active node, wherein the plurality of ports form standby links with a common endpoint; 
 a communication link with an active node; and 
 a switching fabric between the plurality of ports, 
 wherein the standby node is configured to
 remain in a standby state responsive to a loss of the communication link, wherein, in the standby state, all the standby links are non-distributing; 
 monitor frames transmitted by the common endpoint to the standby node over the standby links; and 
 determine based on the monitored frames whether all active links between the active node and the common endpoint have failed and enter an active state with all the standby links distributing based thereon. 
 
 
     
     
       9. The standby node of  claim 8 , wherein the standby node is further configured to
 determine based on the monitoring frames whether less than all of the active links have failed and remain in the standby state and continue monitoring the frames transmitted by the common endpoint over the standby links based thereon. 
 
     
     
       10. The standby node of  claim 8 , wherein the frames are monitored to check for a presence of SYNC bits from the common endpoint with each SYNC bit set to TRUE indicative of a switch by the common endpoint of one of the active links to one of the standby links. 
     
     
       11. The standby node of  claim 8 , wherein the common endpoint is communicatively coupled to both the active node and the standby node in an active/standby triangle topology. 
     
     
       12. The standby node of  claim 8 , wherein the common endpoint is configured to operate Link Aggregation Control Protocol (LACP) and an N:N link-level redundancy between the active node and the standby node. 
     
     
       13. The standby node of  claim 8 , wherein the common endpoint is unaware the active node and the standby node are in separate network elements. 
     
     
       14. The standby node of  claim 8 , wherein the loss of adjacency with the active node is based on a failure or fault on the communication link, while the active node and the standby node are both operational. 
     
     
       15. An apparatus configured for all-or-none switchover to prevent split-brain problems in a Multi-Chassis Link Aggregation Group (MC-LAG) network located at a standby node, the apparatus comprising:
 circuitry configured to remain in a standby state responsive to a loss of adjacency with an active node, wherein, in the standby state, all standby links between the standby node and a common endpoint are non-distributing; 
 circuitry configured to monitor frames transmitted by the common endpoint to the standby node over the standby links; and 
 circuitry configured to determine based on the monitored frames whether all active links between the active node and the common endpoint have failed and enter an active state with all the standby links distributing based thereon. 
 
     
     
       16. The apparatus of  claim 15 , further comprising:
 circuitry configured to determine based on the monitored frames whether less than all of the active links have failed and remain in the standby state and continue monitoring the frames transmitted by the common endpoint over the standby links based thereon. 
 
     
     
       17. The apparatus of  claim 15 , wherein the circuitry configured to monitor checks for a presence of SYNC bits from the common endpoint with each SYNC bit set to TRUE indicative of a switch by the common endpoint of one of the active links to one of the standby links. 
     
     
       18. The apparatus of  claim 15 , wherein the common endpoint is communicatively coupled to both the active node and the standby node in an active/standby triangle topology. 
     
     
       19. The apparatus of  claim 15 , wherein the common endpoint is configured to operate Link Aggregation Control Protocol (LACP) and an N:N link-level redundancy between the active node and the standby node. 
     
     
       20. The apparatus of  claim 15 , wherein the common endpoint is unaware the active node and the standby node are in separate network elements.